Date: Tue, 26 Oct 2021 20:01:44 -0700
From: Noah Misch <noah@leadboat.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Andrey Borodin <x4mmm@yandex-team.ru>,
	PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>,
	Michael Paquier <michael@paquier.xyz>, Peter Geoghegan <pg@bowt.ie>,
	Andres Freund <andres@anarazel.de>
Subject: Re: conchuela timeouts since 2021-10-09 system upgrade
Message-ID: <20211027030144.GA152505@rfd.leadboat.com>
References: <64835CF9-D9AA-49BF-A685-01C23B1023C1@yandex-team.ru>
 <20211024161942.GB3945842@rfd.leadboat.com>
 <640FA6DC-A998-4499-B3B4-2773685B8C56@yandex-team.ru>
 <3792312.1635119599@sss.pgh.pa.us>
 <B5CC4C8F-1B73-4359-AF68-26A08A17BCD1@yandex-team.ru>
 <4170532.1635195582@sss.pgh.pa.us>
 <20211026015157.GA113335@rfd.leadboat.com>
 <52098.1635228234@sss.pgh.pa.us>
 <20211026134500.GA128912@rfd.leadboat.com>
 <139687.1635277318@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <139687.1635277318@sss.pgh.pa.us>
User-Agent: Mutt/1.5.24 (2015-08-30)
Archived-At: 
 <https://www.postgresql.org/message-id/20211027030144.GA152505%40rfd.leadboat.com>
Precedence: bulk

On Tue, Oct 26, 2021 at 03:41:58PM -0400, Tom Lane wrote:
> Noah Misch <noah@leadboat.com> writes:
> > On Tue, Oct 26, 2021 at 02:03:54AM -0400, Tom Lane wrote:
> >> Or more
> >> practically, use advisory locks in that script to enforce that only one
> >> runs at once.
> 
> > The author did try that.
> 
> Ah, I see: if the other pgbench thread is waiting in pg_advisory_lock,
> then it is inside a transaction, so it blocks CIC from completing.
> We can get around that though, by using pg_try_advisory_lock and not
> proceeding if it fails.

Good thought.

> The attached POC does this for the 002 test;
> it looks like the same thing could be done to 003.
> 
> Now the problem with this is it will only work back to v12, because
> pgbench lacks the necessary features before that.  However, I think
> it's worth doing it like this in versions where we can do so, because
> of the load-balancing aspect: this won't waste cycles running CICs
> after the inserts have stopped, nor vice versa.
> 
> Thoughts?

I tried it out with the fix removed (git checkout fdd965d; git checkout HEAD^
src/include src/backend; git apply CIC-test-with-one-pgbench.patch).
Sensitivity (~25%) and runtime (~900ms) were in the same ballpark.  Still,
even if it doesn't buy back notable cycles, the test code seems easier to read
and understand with your change.